Designing Spindle PID
- danimal300
- Offline
- Senior Member
-
Less
More
- Posts: 63
- Thank you received: 1
18 Sep 2017 02:15 - 18 Sep 2017 02:18 #99071
by danimal300
Designing Spindle PID was created by danimal300
I tried this a while back, and could not tune my spindle properly and ended up just deleting it. Now I am in a dynamics and controls class in college and plan to develop a solid closed loop control system for my CNC machine as part of a project. Project guidelines call for full explanation of design choices and how they affect the control, so I think that it would make a good tutorial for future hobby mill configurations similar to mine. I just need some help with the scope and big picture design and I will apply the rest and hopefully document it here.
One big problem that I have is that I am not even sure what production machines are capable of. My entire machining experience has been a self taught run of do's and dont's through hobby mills that have led up to a pretty accurate machine that I feel that I can rely on but want to take to the next level. I want to know that it is loaded properly and removing material as efficiently as possible.
My machine setup: LMS 3990 with Gecko G540 controller, using nema 23 stepper motors. I have a hall effect sensor and 4 magnetic sensor points on the spindle, and so far I have that configured correctly to view RPM as verified by hand held tachometer. I have the spindle control board and can view RPM on my HMI in LinuxCNC.
Mechanical Configuration:
1) I am considering printing a ring and converting the sensor to a IR type sensor with higher resolution than 4 pulses per revolution. But for a machine that has 0 - 2500 RPM capability, how many pulses would be best for optimal resolution and highest accuracy?
2) My spindle is 500W, so it is pretty easy to overload it even if my end mill is well within its chip load. Will I need to consider different material types? Will I need to change PID set points in order to mill steel vs aluminum?
I have read most threads on here for PID, but please feel free to link any and all of them because I tend to overlook information if it does not immediately appear useful to my immediate work. Also feel free to offer any guidance or "what if" suggestions because I want to get the most out of this machine and I have a narrow window to design and build the best controller that I can. I would like to make it as diverse and capable as I can because I may not have time to improve it for a while.
Thanks for all the help in advance!
One big problem that I have is that I am not even sure what production machines are capable of. My entire machining experience has been a self taught run of do's and dont's through hobby mills that have led up to a pretty accurate machine that I feel that I can rely on but want to take to the next level. I want to know that it is loaded properly and removing material as efficiently as possible.
My machine setup: LMS 3990 with Gecko G540 controller, using nema 23 stepper motors. I have a hall effect sensor and 4 magnetic sensor points on the spindle, and so far I have that configured correctly to view RPM as verified by hand held tachometer. I have the spindle control board and can view RPM on my HMI in LinuxCNC.
Mechanical Configuration:
1) I am considering printing a ring and converting the sensor to a IR type sensor with higher resolution than 4 pulses per revolution. But for a machine that has 0 - 2500 RPM capability, how many pulses would be best for optimal resolution and highest accuracy?
2) My spindle is 500W, so it is pretty easy to overload it even if my end mill is well within its chip load. Will I need to consider different material types? Will I need to change PID set points in order to mill steel vs aluminum?
I have read most threads on here for PID, but please feel free to link any and all of them because I tend to overlook information if it does not immediately appear useful to my immediate work. Also feel free to offer any guidance or "what if" suggestions because I want to get the most out of this machine and I have a narrow window to design and build the best controller that I can. I would like to make it as diverse and capable as I can because I may not have time to improve it for a while.
Thanks for all the help in advance!
Last edit: 18 Sep 2017 02:18 by danimal300.
Please Log in or Create an account to join the conversation.
22 Sep 2017 02:10 #99231
by andypugh
Replied by andypugh on topic Designing Spindle PID
If you are using a parallel port then the first thing to work out for the spindle sensor is how fast your parallel port can reliably see all 4 edges of the quadrature counter.
This is a very simple calculation based on the base thread period. Plus some headroom.
This is a very simple calculation based on the base thread period. Plus some headroom.
Please Log in or Create an account to join the conversation.
- danimal300
- Offline
- Senior Member
-
Less
More
- Posts: 63
- Thank you received: 1
23 Nov 2017 05:52 #102246
by danimal300
Replied by danimal300 on topic Designing Spindle PID
I am sure that this is easy to a lot of the folks on here, but I finally got closed loop PID control working on my machine. It is a LMS 3900 benchtop mill that has been converted to CNC and I built a hall effect sensor encoder for the spindle output. I chose a simple detector for the time being to avoid quadrature and simplify tuning.
I need some help optimising the tuning, and see if I can find a way to view the transfer function to a step input. I tried the hal-scope, but did not have much luck. It seemed like the sample period was too small to capture the whole impulse response. I was measuring pid_feedback and hoping to get something that resembles the total overshoot.
I can run the calculations if I can just figure out a way to record the machine response in a way that I can determine the location of the poles in the transfer function.
Thanks for the help!
I need some help optimising the tuning, and see if I can find a way to view the transfer function to a step input. I tried the hal-scope, but did not have much luck. It seemed like the sample period was too small to capture the whole impulse response. I was measuring pid_feedback and hoping to get something that resembles the total overshoot.
I can run the calculations if I can just figure out a way to record the machine response in a way that I can determine the location of the poles in the transfer function.
Thanks for the help!
Please Log in or Create an account to join the conversation.
- danimal300
- Offline
- Senior Member
-
Less
More
- Posts: 63
- Thank you received: 1
23 Nov 2017 06:08 #102247
by danimal300
Replied by danimal300 on topic Designing Spindle PID
Here is my .hal file, can I get some feedback on general efficiency and best practices? I am getting better at organizing everything, but I still have a lot to learn.
loadrt trivkins
loadrt [EMCMOT]EMCMOT base_period_nsec=[EMCMOT]BASE_PERIOD servo_period_nsec=[EMCMOT]SERVO_PERIOD num_joints=[TRAJ]AXES
# Parallel port setup
loadrt probe_parport
loadrt hal_parport cfg="0x378 out"
setp parport.0.reset-time 1000
addf parport.0.read base-thread
addf parport.0.write base-thread
addf parport.0.reset base-thread
net spindle-ccw => parport.0.pin-01-out
net xstep => parport.0.pin-02-out
setp parport.0.pin-02-out-reset 1
setp parport.0.pin-03-out-invert 1
net xdir => parport.0.pin-03-out
net ystep => parport.0.pin-04-out
setp parport.0.pin-04-out-reset 1
net ydir => parport.0.pin-05-out
net zstep => parport.0.pin-06-out
setp parport.0.pin-06-out-reset 1
net zdir => parport.0.pin-07-out
net astep => parport.0.pin-08-out
setp parport.0.pin-08-out-reset 1
net adir => parport.0.pin-09-out
net charge-pump => parport.0.pin-16-out
net spindle-on => parport.0.pin-17-out
net spindle-phase-a <= parport.0.pin-10-in
net spindle-index <= parport.0.pin-11-in
# Stepgen setup
loadrt stepgen step_type=0,0,0,0
addf stepgen.make-pulses base-thread
loadrt abs count=1
addf stepgen.capture-position servo-thread
addf stepgen.update-freq servo-thread
# Initialize the scale (This is used to scale the encoder pulse to RPM)
# Scaling factor 6 pulses per revolution x 60 sec per min = 360
loadrt scale count=3
addf scale.0 servo-thread
setp scale.0.gain 360
addf scale.1 servo-thread
setp scale.1.gain 60
addf scale.2 servo-thread
setp scale.2.gain 60
# Initilize Low Pass Filter for Spindle
loadrt lowpass count=2
addf lowpass.0 servo-thread
addf abs.0 servo-thread
addf lowpass.1 servo-thread
# E-Stop and Charge Pump enable
loadrt charge_pump
net estop-out charge-pump.enable iocontrol.0.user-enable-out
net charge-pump <= charge-pump.out
addf charge-pump base-thread
net estop-out <= iocontrol.0.user-enable-out
net estop-out => iocontrol.0.emc-enable-in
loadrt near
addf near.0 servo-thread
# Initialize the encoder (spindle)
loadrt encoder num_chan=1
addf encoder.update-counters base-thread
addf encoder.capture-position servo-thread
setp encoder.0.position-scale 6.000000
setp encoder.0.counter-mode 1
# Initialize the PID
loadrt pid num_chan=1
addf pid.0.do-pid-calcs servo-thread
setp pid.0.Pgain 0.003
setp pid.0.Igain 1
setp pid.0.Dgain 0.004
setp pid.0.maxoutput 80
# Initialize the pwmgen
# loadrt pwmgen output_type=1
loadrt pwmgen output_type=0
addf pwmgen.update servo-thread
addf pwmgen.make-pulses base-thread
setp pwmgen.0.pwm-freq 50.0
setp pwmgen.0.scale 4000.0
setp pwmgen.0.offset -.0090
setp pwmgen.0.dither-pwm true
#setp pwmgen.0.min-dc 0.001 # min output duty cycle --> lowest possible spindle speed
#setp pwmgen.0.max-dc .8 # max output duty cycle --> highest possible spindle speed
# Initialize the sum2
loadrt sum2 count=1
addf sum2.0 servo-thread
# Initialize the inverter for negative feedback
loadrt invert count=1
addf invert.0 servo-thread
addf motion-command-handler servo-thread
addf motion-controller servo-thread
# Spindle PID closed loop controler
net spindle-cw <= motion.spindle-forward
net spindle-phase-a encoder.0.phase-A
net spindle-index encoder.0.phase-Z
net spindle-on motion.spindle-on => pwmgen.0.enable => pid.0.enable
net spindle-pwm pwmgen.0.pwm => parport.0.pin-14-out
net spindle-index-enable encoder.0.index-enable <=> motion.spindle-index-enable
net spindle-position encoder.0.position => motion.spindle-revs
net spindle-velocity encoder.0.velocity => scale.2.in
net pid_feedback pid.0.feedback <= motion.spindle-speed-in <= scale.2.out
net pid_command pid.0.command <= motion.spindle-speed-out => sum2.0.in1
net pid_error pid.0.error
net pid_output pid.0.output => sum2.0.in0
net pwmgen_value pwmgen.0.value <= sum2.0.out
setp stepgen.0.position-scale [AXIS_0]SCALE
setp stepgen.0.steplen 2500
setp stepgen.0.stepspace 2500
setp stepgen.0.dirhold 1000
setp stepgen.0.dirsetup 1000
setp stepgen.0.maxaccel [AXIS_0]STEPGEN_MAXACCEL
net xpos-cmd axis.0.motor-pos-cmd => stepgen.0.position-cmd
net xpos-fb stepgen.0.position-fb => axis.0.motor-pos-fb
net xstep <= stepgen.0.step
net xdir <= stepgen.0.dir
net xenable axis.0.amp-enable-out => stepgen.0.enable
setp stepgen.1.position-scale [AXIS_1]SCALE
setp stepgen.1.steplen 2500
setp stepgen.1.stepspace 2500
setp stepgen.1.dirhold 1000
setp stepgen.1.dirsetup 1000
setp stepgen.1.maxaccel [AXIS_1]STEPGEN_MAXACCEL
net ypos-cmd axis.1.motor-pos-cmd => stepgen.1.position-cmd
net ypos-fb stepgen.1.position-fb => axis.1.motor-pos-fb
net ystep <= stepgen.1.step
net ydir <= stepgen.1.dir
net yenable axis.1.amp-enable-out => stepgen.1.enable
setp stepgen.2.position-scale [AXIS_2]SCALE
setp stepgen.2.steplen 2500
setp stepgen.2.stepspace 2500
setp stepgen.2.dirhold 1000
setp stepgen.2.dirsetup 1000
setp stepgen.2.maxaccel [AXIS_2]STEPGEN_MAXACCEL
net zpos-cmd axis.2.motor-pos-cmd => stepgen.2.position-cmd
net zpos-fb stepgen.2.position-fb => axis.2.motor-pos-fb
net zstep <= stepgen.2.step
net zdir <= stepgen.2.dir
net zenable axis.2.amp-enable-out => stepgen.2.enable
setp stepgen.3.position-scale [AXIS_3]SCALE
setp stepgen.3.steplen 1
setp stepgen.3.stepspace 0
setp stepgen.3.dirhold 20200
setp stepgen.3.dirsetup 20200
setp stepgen.3.maxaccel [AXIS_3]STEPGEN_MAXACCEL
net apos-cmd axis.3.motor-pos-cmd => stepgen.3.position-cmd
net apos-fb stepgen.3.position-fb => axis.3.motor-pos-fb
net astep <= stepgen.3.step
net adir <= stepgen.3.dir
net aenable axis.3.amp-enable-out => stepgen.3.enable
loadusr -W hal_manualtoolchange
net tool-change iocontrol.0.tool-change => hal_manualtoolchange.change
net tool-changed iocontrol.0.tool-changed <= hal_manualtoolchange.changed
net tool-number iocontrol.0.tool-prep-number => hal_manualtoolchange.number
net tool-prepare-loopback iocontrol.0.tool-prepare => iocontrol.0.tool-preparedPlease Log in or Create an account to join the conversation.
- danimal300
- Offline
- Senior Member
-
Less
More
- Posts: 63
- Thank you received: 1
23 Nov 2017 06:28 #102248
by danimal300
Replied by danimal300 on topic Designing Spindle PID
My Encoder design:
Here are the cad files for the sensor mount. It fits on the LMS 3990 mill, not sure if it will work on others.
Spindle Lock Cap
Main Sensor Body
Sensor Ring
Here are the cad files for the sensor mount. It fits on the LMS 3990 mill, not sure if it will work on others.
Spindle Lock Cap
Main Sensor Body
Sensor Ring
Please Log in or Create an account to join the conversation.
23 Nov 2017 12:26 #102256
by andypugh
You can change the Haslcope sampling frequency to record for longer. By default it samples every 1mS but you can sample every 2, every 4 and so on.
Click where is says something like "4000 samples in 4 seconds"
I think that you might be over-thinking this, with the talk of poles etc.
Replied by andypugh on topic Designing Spindle PID
I need some help optimising the tuning, and see if I can find a way to view the transfer function to a step input. I tried the hal-scope, but did not have much luck. It seemed like the sample period was too small to capture the whole impulse response.
You can change the Haslcope sampling frequency to record for longer. By default it samples every 1mS but you can sample every 2, every 4 and so on.
Click where is says something like "4000 samples in 4 seconds"
I think that you might be over-thinking this, with the talk of poles etc.
Please Log in or Create an account to join the conversation.
- danimal300
- Offline
- Senior Member
-
Less
More
- Posts: 63
- Thank you received: 1
23 Nov 2017 18:32 - 26 Nov 2017 22:16 #102268
by danimal300
Replied by danimal300 on topic Designing Spindle PID
I am over thinking the design of the controller, but that is because it is for a college dynamics and controls class. I would really like to dive into the inner workings and really tune the response of the controller, so I just need a way to determine the transfer function of the plant and I can use matlab sisotool to really tweak the inputs. I have a perfectly functional controller, but the fine tuning aspects are what I am shooting for with this project.
Here is a video of it functioning. It is a little long because I plan on clipping it down and creating a project video for the class presentation:
Here is a video of it functioning. It is a little long because I plan on clipping it down and creating a project video for the class presentation:
Last edit: 26 Nov 2017 22:16 by danimal300. Reason: Changed Video Link
Please Log in or Create an account to join the conversation.
- danimal300
- Offline
- Senior Member
-
Less
More
- Posts: 63
- Thank you received: 1
23 Nov 2017 18:36 #102269
by danimal300
Replied by danimal300 on topic Designing Spindle PID
Is there any way to sample pins and dump the information into a spreadsheet? I see the hal histogram, and will play with that a little bit but it looks a little limited, and it would be nice to select a couple pins and have it dump some values into a spreadsheet on like a servo thread periodicity.
Any ideas on how I could get this to print out?
Any ideas on how I could get this to print out?
Please Log in or Create an account to join the conversation.
23 Nov 2017 19:26 #102270
by andypugh
Replied by andypugh on topic Designing Spindle PID
You can save the data file from Halscope.
The following user(s) said Thank You: danimal300
Please Log in or Create an account to join the conversation.
23 Nov 2017 21:12 #102274
by Muzzer
Replied by Muzzer on topic Designing Spindle PID
A machine technician would traditionally start with P only, increasing until instability starts, then add some D to improve the transient response and some I to eliminate any offset. Of course, they all interact, so it tends to be an iterative process. It helps if you are watching the output (speed) so you can observe any overshoot etc. With some messing about you should be able to get a reasonable result. I'm sure there are loads of Pootube videos out there that cover this.
A slightly more insightful approach would be to run the system open loop and give it a step input (set voltage). Then observe the delay (s) and response rate (rpm per second etc) and use a rule of thumb method like Ziegler Nichols to estimate some starting values.
But if you are aiming to become a control engineer in some way (ie what you should be doing here?), you should be making your own model of the plant - it's not difficult. For that you simply need stuff like the moment of inertia of the spinning bits (taking account of the gear ratios etc), the volts per rev of the feedback signal and the torque-volts gain of the motor vs the control voltage etc. This is fairly simple stuff and I expect you will probably find something very close to what you need if you check out the Matlab tutorials.
From the system you have, you need to be able to see the demand signal (presumably a DC voltage to the spindle VFD) and the feedback signal from the pickup, ideally normalised to the same basis as the demand signal. It's the difference between these 2 signals that the PID controller works on.
Murray
A slightly more insightful approach would be to run the system open loop and give it a step input (set voltage). Then observe the delay (s) and response rate (rpm per second etc) and use a rule of thumb method like Ziegler Nichols to estimate some starting values.
But if you are aiming to become a control engineer in some way (ie what you should be doing here?), you should be making your own model of the plant - it's not difficult. For that you simply need stuff like the moment of inertia of the spinning bits (taking account of the gear ratios etc), the volts per rev of the feedback signal and the torque-volts gain of the motor vs the control voltage etc. This is fairly simple stuff and I expect you will probably find something very close to what you need if you check out the Matlab tutorials.
From the system you have, you need to be able to see the demand signal (presumably a DC voltage to the spindle VFD) and the feedback signal from the pickup, ideally normalised to the same basis as the demand signal. It's the difference between these 2 signals that the PID controller works on.
Murray
Please Log in or Create an account to join the conversation.
Time to create page: 0.175 seconds